Rapid construction of 4D high-quality microstructural image for cement hydration using partial information registration. (April 2022)

• Record Type: Journal Article
• Authors: Zhang, Liangliang; Wang, Lin; Yang, Bo; Niu, Sijie; Han, Yamin; Oh, Sung-Kwun
• Is Part Of: Pattern recognition. Volume 124(2022)
• Publication Date: 2022-04
• Subjects: Cement hydration -- Rapid image construction -- Image registration -- Particle swarm optimization -- Microstructural temporal image sequences; Pattern perception -- Periodicals; Perception des structures -- Périodiques; Patroonherkenning; 006.4
• DOI: 10.1016/j.patcog.2021.108471
• Languages: English
• ISSNs: 0031-3203

Abstract

Highlights: A partial information registration (PIR) method is proposed to speed up registration and to improve registration accuracy by adopting partial information on the spatial and phased scales, respectively. A multi-factor multi-layer particle swarm optimization (MFML-PSO) algorithm is designed to improve the optimization performance of particle swarm optimization in multimodal problems. An image processing framework is established to rapidly construct four-dimensional highquality microstructural images for cement hydration. Fast bias field correction and temporal intensity calibration are implemented to enhance the quality of microtomography temporal image sequences for cement hydration. Experimental results indicate that the four-dimensional high quality microstructural image can be constructed rapidly for cement hydration with promising precision. Abstract: Studying on the microstructural evolution of cement paste during hydration is of considerable significance for understanding its mechanism and designing such material in cement industry. With the use of microtomography and image registration, the four-dimensional (4D) microstructure of cement paste can be captured, thereby assisting material scientists in studying the hydration process in situ . However, as a challenging task, the construction of high-quality 4D microstructural image is remarkably impeded by image size, isotropy, and homogeneity. This paper proposes an image processing framework to construct 4D high quality microstructural image rapidly for cement hydration. This framework improves and accelerates microstructural image registration and enhancement by using bias field correction, temporal intensity calibration and fast image registration. Additionally, a partial information registration method adopting partial information on the spatial and phased scales, is proposed to improve the registration speed and accuracy. Furthermore, a multi-factor multi-layer particle swarm optimization is proposed to improve the optimization in registration. Experimental results indicate that the 4D high quality microstructural image can be constructed rapidly with promising precision.